Recent increases in age-specific survival among the elderly are expected to generate enormous demands for health care. Predictions of the extent of these demands, however are hampered by a death of empirical data on time- trends in age-specific morbidity and the impact of these changes on utilization. A valuable opportunity to investigate such trends exists with the unique medical records linkage system of the Rochester Epidemiology Project (REP) and the demonstrated capacity for population-based epidemiologic and health services research in Olmsted County, MN. Over the past 2 decades Olmsted County has mirrored the U.S. in that mortality rates have declined for persons > 65 years of age, while hospital use rates have increased, especially among the oldest old. These observations provide provisional support for the argument that greater demands for health care services will accompany increased longevity as a result of higher levels of age-specific morbidity. This proposal begins to tap the potential of the REP for testing this hypothesis by focusing specifically on individuals characterized by the chronic disease, non-insulin dependent diabetes mellitus (NIDDM). This retrospective, longitudinal study will compare 3 prevalence cohorts consisting of all Rochester, MN, residents > 45 with a diagnosis of NIDDM as of January for the years 1970, 1980, 1990. Individuals will be followed from data of prevalence until 1/1/94. Differences among cohorts for age-specific survival and duration of NIDDM will be estimated, as will the association between survival and 2 distinct measures of morbidity: (1) incidence of comorbidity following date of prevalence, and (2) a summary measure of morbidity defined as overall disease severity. Cohorts will be compared for risk of hospitalization and hospitalization incidence density. Disease duration, morbidity, and hospital use rates will be determined with review of patient-based medical records (outpatient and inpatient). Multiple regression and Cox proportional hazards models will be employed to determine trends in the independent association between overall morbidity and hospital utilization. Extensive experience in REP studies of numerous diseases verifies the ability to determine appropriate clinical and epidemiologic parameters and to provide nearly 100% follow-up to 1/1/94 for each of the prevalence cohorts. This retrospective study of clinically defined disease and hospitalization rates will also provide a foundation for future prospective studies of active life expectancy and long term care use.